RE: [xml-dev] Composability: Do the Emperor's Clothes Fit? (RE: [xml-dev] Costs, benefits, and emperors with wardrobe malfunctions - wasRe: [xml-dev] WS-Emperor naked?)

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

It's a conversation with a long history.  :-)

http://groups.yahoo.com/group/xml-dev/message/23220
http://lists.xml.org/archives/xml-dev/200309/msg00199.html
http://lists.xml.org/archives/xml-dev/200009/msg00081.html
http://aspn.activestate.com/ASPN/Mail/Message/xml-dev/1052067
http://lists.xml.org/archives/xml-dev/200202/msg00305.html

By vette, I mean, what conformance tests are being applied?
 
The Concept of Operations will have to be fairly detailed 
and if we are hooking up blindly, we have to trust that 
this has been done.  Given mission, risk costs go up.  That 
suggests to me multiple registries because the communities 
QoS requirements will definitely send up costs.  A single 
registry won't scale and multiple registries don't have to 
(all politics is local).   

The registry concepts are abstract.  For this 
to work for justice systems, a high QoS number will 
be set for some services.   Now performance counts and 
if decisions made on the standard data dictionary cause 
those numbers to fall, the data dictionary loses.  Therein 
is the rub of grand plans ahead of implementations.  We've 
all been there (See CALS).  :-)

SOAs as standard across an industy may be devilishly 
hard to achieve.  Not impossible, but not timely. 
Composition can't be just-in-time for some industries, 
but I'm preaching to the choir.

len


From: Chiusano Joseph [mailto:chiusano_joseph@bah.com]

"Bullard, Claude L (Len)" wrote:
> 
> Please explain how a registry solves the problem
> of including an insecure or unreliable software
> component to process a given part of the composition.

Sure - it could solve it by holding the registry authorities accountable
for testing and "certifying" the reliability of a given software
component, according to criteria that are created as part of the concept
of operations (CONOPS) for the registry, perhaps by the Community of
Interest (CoI) that the registry serves.
 
> Composability is based on namespaces.   What is
> registered, the namespace or the implementation?

Ah, that brings me to a proposal for a "Namespace Manager" function that
I offered to the ebXML Registry TC back in January 2002:

http://lists.oasis-open.org/archives/regrep/200201/msg00061.html

More than 2 years later, I still very much see the need for such a
function as "native" to the registry architecture - we've laid some
foundations in the meantime that have brought us closer to realizing
this goal. I hope this feature will be included in a future version of
the registry specs.
 
> Does the registry authority vette the implementation?

Depends on what you mean by "vette", and with whom.

> How would that be sustained?
> 
> Does the implementor check the registry or does
> the RADE vendor pass a conformance test and then
> is enabled to register the component such that
> selection of a service within the RADE is warrantied?

It all depends on the registry CONOPs and policies.
 
> Would you conjecture that Homeland Security systems
> such as those that support GJXDM will require
> registration and vetting?

The GJXDM folks have been talking about registries for quite some time,
so I foresee that in the future. They are also now discussing SOAs, so
registries will become even more important than ever.